Nearly a decade after research firms predicted major cost savings and clinical benefits from the use of health-IT, adoption rates among U.S. medical providers remain sluggish, with the industry slow to embrace the big-data movement.
Electronic health record (EHR) adoption has been fastest at larger, more technology-savvy medical organizations, while smaller practitioners -- which make up the bulk of U.S. clinicians -- have been slow on the health-IT uptake for a variety of reasons, chief among them the cost, but also the training time and effort needed to make the move from paper. Those that don't adopt EHRs by 2015 face decreased government reimbursements for Medicare patients under existing law.
[ IN PICTURES: High-tech healthcare technology gone wild ]
Questions surrounding the effectiveness and financial impact of EHRs were raised in a January report from RAND Corp., which reconsidered its 2005 prediction that health-IT, including EHRs, could save the U.S. approximately US$81 billion annually. The new report on the technology noted that health costs have increased since the 2005 research and called the clinical benefits "mixed."
But there are EHR success stories such as Kaiser Permanente and the Cleveland Clinic, and some health-care providers are rapidly embracing the benefits that big-data analytics brings to clinical medicine.
"The government is pushing [EHR use] along and it's becoming a standard of care," said Tom Handler, research director of Gartner's health-care provider analyst group and a former physician. "At some point in the near future, not heading down the path of an EHR [system] will not serve an organization well."
Long-term use of health-IT can lead to reduced care costs, even if expenses initially increase, he said.
In addition to the cost of the EHR system itself, its use could also increase medical costs by reminding doctors to order required tests that they forgot, he said. But this detailed care should lead to healthier patients who need less expensive medical treatments overall. For instance, an EHR system that reminds a doctor to contact a diabetic patient to schedule an appointment could lead to lower long-term care costs if the software ultimately helps to prevent the need for the patient to have a foot amputated, he said.
The U.S. government is nudging providers to adopt health-IT and is picking up some of the tab for EHR deployments. Costs for a system vary widely, from tens of thousands of dollars for a small practice to upwards of $1 billion for a major system implementation.
As part of the American Recovery and Reinvestment Act, the government dedicated $20.6 billion to EHR projects. Providers must show that EHR use improved health care through a series of reimbursement guidelines called meaningful use. The first phase dealt with purchasing and rolling out EHR systems and began in 2011. The second phase includes data interoperability provisions and started this year. Phase three covers improving outcomes and starts in 2015.
EHR use is greatest at multidisciplinary practices with more than 50 doctors, Handler says. At smaller clinics the adoption rate is between 5 percent and 10 percent, he says, adding that 75 percent of U.S. physicians practice in groups of five or fewer doctors.
As smaller practices merge, Handler sees the EHR adoption rate rising, but "100 percent EHR adoption could be another decade."
Jed Weissberg, Kaiser Permanente
Most providers are familiarizing themselves with EHRs but only a few organizations are delving into data projects, said Jed Weissberg, senior vice president, hospitals, quality and care delivery excellence and a doctor at Kaiser Permanente, a 9 million-member health plan and care provider in Oakland, California. "The rest of American medicine is getting on the platform and getting up to speed."
But incorporating EHRs into health care does not automatically translate into robust data analysis. IT vendors, hospital executives and physicians pose unique challenges to getting big data into medicine.
Core EHRs "are still mainly used for documentation of individual patient visits and don't contain all the data you might like," Weissberg said.
At Kaiser, which completed rolling out its Epic Systems-powered EHR system, Kaiser Permanente HealthConnect, to its 37 hospitals and 533 medical offices in 2010, EHRs lacked the ability to capture pain and functionality data on patients who had joint surgery, for instance. Post-surgery metrics could be placed in a database that could help potential joint surgery patients learn how people with similar clinical conditions benefited from the operation.
Kaiser will start capturing this data and is looking into having patients enter this information via a secure Web portal or having a physician assistant handle the task with a tablet during the follow-up visit.
"Everybody wants this approach to big data and more advanced analytics to work," Weissberg said, noting that querying a database to research unique medical conditions is cheaper and faster than launching a research study.
Personalized medicine usually entails talk of genomics, but EHRs and data mining can also lead to tailored care, he said.
"Starting with overall population we can define subsets by age, gender, clinical conditions and body mass index that are enough like you. We have enough data that we can at least show what happened to them."
Ramy Arnaout, Beth Israel Deaconess Medical Center
For now, EHR functionality is limited to what developers build into the software, said Ramy Arnaout, a pathologist and associate director of the clinical microbiology laboratory at Beth Israel Deaconess Medical Center in Boston, noting that his views are his own and don't necessarily reflect those of his employer.
"If they don't build software that lets you mine data in the kind of ways that you're suggesting then you're largely out of luck," he said.
Vendors lack a financial incentive to add this functionality to their EHRs since they're already making profits on applications without these abilities. Adding data analysis functions means creating systems with greater interoperability and that threatens vendor revenue since it makes moving to another EHR system easier, he said.
But interoperability, which entails feeding standardized data that comes from a hospital's many departments into a single analysis platform, is essential for big data, said Weissberg.
"To get at that notion of big data, you need to take information from multiple data marts and systems that are very disparate and aren't built in an integrated way."
Hospital executives may not be eager to customize their EHR systems with data analysis modules since such additions can knock a customer off a vendor's standard upgrade cycle and result in multimillion-dollar upgrades, said Forrester vice president and principal analyst Craig Le Clair.
Instead, hospitals are turning to agile technologies, like mobile applications and tablets, to plug into their EHRs. These technologies are dedicated to the "$50 billion problems that we have in health care," like analyzing readmissions, he said.
Executives are also watching the bottom line and won't fund IT projects that lack a proven return on investment. Large data-analysis projects fall under this category, said Arnaout.
"You're asking folks, when they have no money, to take a chance on something that is unproven when they've got a lot of other things that are less sexy, but are proven," he said. "Right or wrong, unless you have a case study showing an ROI they're going to be watching their dollars."
Doctors too harbor doubts on data analysis' benefits although how they practice medicine -- analyzing data stored over time from a variety of sources and constantly updating it to develop an output based on certain parameters -- is how EHR data analysis would also function. Computers differ from humans in their ability to take all the data on a topic, keep it in memory indefinitely and generate outcome percentages with confidence values, Arnaout said.
"Providers don't think of going to a big-data-type interface on the computer because they haven't seen it even though they're that computer right now." And with doctors focused on treating patients they're not demanding this technology, he added.
Building systems that medical professionals can trust requires hiring people with a strong understanding of medicine and computer science. Doctors with coding skills may not write the program but knowledge of both fields can lead to software that generates clinical value, said Arnaout.
With software developers in strong demand across many industries, programmers are accepting positions at technology companies with lucrative salaries that exceed the compensation hospitals offer, he said.
"We may be coming out of a recession but there's never been a recession for coders for the past 10 years."
Cloud-based EHR data analytics services have started to enter the market, touting benefits like lower IT costs and access to medical information from multiple health-care systems.
Cleveland Clinic is among the providers well-poised to take advantage of those technological advances.
Credit: Fred O'Connor
EHRs and data analysis play important roles in the Cleveland Clinic's future clinical care plans, said Chris Coburn, who leads the hospital's venture arm.
"As we try to create the Cleveland Clinic of the future that enables us to best help our patients and also be as financially viable as possible, the use of EHRs and now with these big-data tolls that are arriving, it's a perfect match," said Chris Coburn, executive director of Cleveland Clinic Innovations, the corporate venture arm of the Cleveland Clinic, which records 5.1 million patient visits annually and began using EHRs from Epic Systems in the late 1990s.
In 2009 the Cleveland Clinic launched Explorys, which sells a private SaaS (software-as-a-service) platform that allows medical professionals to explore care options using clinical, financial and operational data from 120 hospitals and 15 million patients. At the time, security concerns were paramount because of the sensitive nature of the data.
"In a relatively short period of time a very secure system was created that we all trust now," Coburn said. "We all worry about this. We're all patients. Folks will have as much confidence in security for their electronic medical records as it relates to things like banking electronically. People treat HIPAA extremely seriously." (The Health Insurance Portability and Accountability Act -- HIPAA -- was passed in 1996 and it governs, among other things, medical-data privacy.)
For hospitals, cloud systems like Explorys offer affordable access to troves of data compared to the expense of developing their own systems that will contain less information, said Colburn. "We're not there yet in terms of where big data is going to take health care. But we are moving there pretty rapidly."
Last Wednesday, the American Society of Clinical Oncology (ASCO) debuted the prototype for its CancerLinQ database. The prototype, which focuses on breast cancer, contains deidentified information from EHRs, care providers and researchers. Physicians will be able to use the data stored in CancerLinQ's full build, which will be a private cloud accessible via a Web portal, to develop treatment plans tailored to a patient's specific cancer and clinical condition.
Credit: Fred O'Connor
Clifford Hudis, president elect of the American Society of Clinical Oncology, discusses the CancerLinQ prototype last week.
"The CancerLinQ prototype leverages a number of new IT trends: the availability of low-cost storage, the affordability and rapid scalability of virtual (cloud) servers, the growth and maturity of open-source software, as well as NoSQL (unstructured) databases," said Dr. Clifford Hudis, president-elect of ASCO, in an email interview. "Many of the benefits from these trends will likely carry forward into the full build."
The final build will include natural language processing, machine learning algorithms and distributed computing, among other technologies, he said. The final role of open-source software is still being considered, although open-source applications were mostly used in the prototype, according to an ASCO webcast. Hadoop, an open-source program used to distribute data processing loads, is a staple in large-scale data analysis.
Because the medical community is slow to incorporate IT into its workflow, these technologies were selected to make the adoption process easier.
"Typically, health care is slow to adopt IT due to the high cost associated with implementations and competing priorities," Hudis said. "Many of our goals around the architecture chosen was to reduce implementation burdens for practices and physicians."
With 85 percent to 90 percent of ASCO members using EHRs, oncology seems like a good match for a large data-analysis project, Hudis said. Additionally, patients and providers are very willing to volunteer their health data, he added. ASCO, whose 30,000 members are physicians and health care professionals representing all fields of cancer treatment and research, received 130,000 EHRs to populate the prototype after initially setting a goal of 30,000 patients.
Regardless of government mandates on EHR use, data analysis will lead to tangible patient benefits.
"We've got all this stuff in the health care guidelines and things that are reimbursable and they end up being a little bit divorced from how our patients actually end up feeling," said Beth Israel's Arnaout. "[Data analytics] is actually going to help how our patients actually end up feeling."